Synergistic interaction of Re complex and amine functionalized multiple ligands in metal-organic frameworks for conversion of carbon dioxide

A metal-organic framework (MOF) is composed of secondary building units (SBUs) of metal ions and organic ligands to link each SBU. Moreover, the photosynthetic synthesis of a valuable CO chemical from carbon dioxide (CO2) represents an important class of appealing methods. Herein, we find that a molecular photocatalyst with high selectivity and activity can be designed via a fine balance in the proximity of Re complex (ReI(CO)3(BPYDC)(Cl), BPYDC = 2,2′-bipyridine-5,5′-dicarboxylate) and -NH2 functionalized multiple ligands composing a MOF photocatalyst, denoted as Re-MOF-NH2. These ligands in Re-MOF-NH2 has been confirmed by infrared, UV-visible, and 1H nuclear magnetic resonance spectra. Moreover, we show from extended X-ray absorption fine structure and in-situ infrared spectra that the bond corresponding to Re-CO upon introduction of -NH2 functional groups is divided into asymmetric bonds of 1.4 Å and 2.3 Å along with different CO2 vibrations, thus making the configuration of carbonyl groups in a Re metal complex become asymmetric in addition to aiding formation of CO2 intermediates within Re-MOF-NH2. Indeed, both of the uneven electron distribution in asymmetric carbonyl groups for Re-CO and the intermolecular stabilization of carbamate intermediates are proven to give the approximately 3-fold increase in photocatalytic activity for conversion of CO2 into CO.

ligand centres often does not have functional group enabling cooperation conversion of carbon dioxide into CO with high selectivity and activity. Additionally, absorption of visible light on the main solar energy spectrum is still limited for even molecular catalysts with multiple ligands. Therefore, it is advantageous for a molecular catalyst to have conjugated ligands enabling synergistic conversion of carbon dioxide upon visible light absorption.
Herein, we report a new MOF with multiple functional ligands enabling structural conjugation for photocatalytic conversion of CO 2 , where a fine balance in the proximity of two functional groups inside the molecular MOF catalyst is proven to result in cooperative photocatalytic activity for carbon dioxide conversion under visible light. We chose the amine (-NH 2 ) functional group as the chemical functionality for incorporation into Re-MOF [9][10][11][12][13] with ReI(CO) 3 (BPYDC)(Cl), BPYDC = 2,2′-bipyridine-5,5′-dicarboxylate, hereafter referred to as ReTC, which is covalently attached within a zirconium MOF based on a UiO-67-type structure 14 . The -NH 2 functionalized Re-MOF having a controlled amount of -NH 2 groups in mol% of X to the total amount of organic linkers is also denoted as Re-MOF-NH 2 (X%) and Fig. 1 shows that the -NH 2 functional groups contribute to make the molecular structure of ReTC asymmetric while it helps to form the CO 2 intermediate within Re-MOF-NH 2 (33%). MOFs are doped with [9][10][11][12][13] , built by one kind [15][16][17][18][19][20][21] of molecular ligands, and/or combined with inorganic nanomaterials 9,12,13,22 to give varying levels of photocatalytic activities. Our work has been concentrated on realizing the molecular photocatalysts capable of giving unique cooperative functionality through structural conjugation of two different ligand centres within MOFs. Moreover, the combined experimental methods using inductively coupled plasma atomic emission spectroscopy (ICP-AES) analysis, X-ray diffraction (XRD) patterns, extended X-ray absorption fine structure (EXAFS) spectra, ultraviolet-visible (UV-vis) and in-situ infrared (IR) spectra, X-ray photoelectron spectroscopy (XPS) scanning electron microscope (SEM) and tunnelling electron microscope (TEM) images, and gas chromatograph data have been also used to probe the cooperative photocatalytic conversion mechanism of carbon dioxide on Re-MOF-NH 2 having the conjugated ReTC and amine functionalized ligand centres.

Results and Discussion
Sample preparation. The Re-MOF-NH 2 (X%) samples were prepared by varying the ratio of 2,2′-diami- Fig. S1 in the Supplementary Information] in the range of 0 mol% (-NH 2 -free Re-MOF) to 80 mol% (-NH 2 at the maximal loading), where the experiment was carried out by increasing the amount of H 2 BPDC-(NH 2 ) 2 by 20, 40, 60, and 80 mol% compared to the total amount of H 2 ReTC and 4,4′-biphenyldicarboxylic acid (H 2 BPDC). The optimal amount of H 2 ReTC was determined to be 20 mol% in the preparation solution of Re-MOF-NH 2 (X%) for CO 2 conversion. The amount of H 2 ReTC (Fig. 1a) depends on the ratio of H 2 BPDC-(NH 2 ) 2 . The protonated combinations of these three linkers and ZrCl 4 were dissolved in a mixed solution of DMF/acetic acid in a 20-mL screw-capped vial, which were then heated at 85 °C for 12 hours. The orange suspensions produced in these processes were collected and washed three times with DMF using a centrifuge (8000 rpm for 10 minutes) and sonication. Then, it has been sequentially immersed in methanol for three 24-hour periods. Finally, the samples were activated by removing the solvent under vacuum for 12 hours at room temperature. These processes were found to result in Re-MOF-NH 2 (X%) with X = 0, 33, 52, 68, and 80, which were identified (Fig. 2a) by the combined experiments using the inductively coupled plasma atomic emission spectroscopy (ICP-AES) and the 1 H NMR spectroscopy of a digested solution of these samples ( Fig. S2 and Table S1 in the Supplementary Information).
Structural characterization of Re-MOF-NH 2 (X%). The crystallinity of Re-MOF-NH 2 (X%) was determined by the PXRD analysis (Figs 2b and S3 in the Supplementary Information), which gave sharp diffraction lines matching with those of the simulated pattern obtained from the experimental X-ray crystal diffraction data of Re-MOF. This gives a clear evidence for the preservation of the Re-MOF structural arrangement upon introduction of -NH 2 functional groups at varied ratios and it is notable that all the samples were synthesized as nanoparticle morphologies. The representative scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images (Figs 2c and S4 in the Supplementary Information) of Re-MOF-NH 2 (X%) demonstrate further that the size of the particles is approximately 300 nm and particles in all the samples have octahedral geometries, regardless of the amount of -NH 2 functional groups incorporated. The permanent porosity of all the samples was found to be preserved, as confirmed by measurement of the N 2 adsorption isotherms (Figs 2d and S5 in the Supplementary Information). The results showed Type I isotherms similar to those of Re-MOFs and UiO-67 9, 14 . The Langmuir surface areas for Re-MOF, Re-MOF-NH 2 (33%), Re-MOF-NH 2 (52%), Re-MOF-NH 2 (68%), and Re-MOF-NH 2 (80%) were 2049, 1900, 1845, 1641, and 1500 m 2 g −1 , respectively. This difference in the surface areas was attributed to the amount of the -NH 2 functional groups and the cramped pore structure in the presence of mixed functional groups. Photocatalytic activity and stability of Re-MOF-NH 2 (X%). All Re-MOF-NH 2 (X%) samples were explored for photocatalytic CO 2 conversion rate under visible light. The samples, which were kept in vacuum overnight, were placed in a stainless-steel reactor and purged with CO 2 and the conversion was initiated in the presence of triethylamine (TEA) under visible light (400-700 nm) using a 300 W Xenon arc lamp. The products produced through the photocatalytic reaction were analysed and quantified using the gas chromatography (GC) equipped with a flame ionization detector (FID) and normalized to the mass of the sample. Figure 3a shows photocatalytic activities of the Re-MOF-NH 2 (X%) samples. The results show that the photocatalytic conversion is the highest when 33 mol% of -NH 2 functional groups, denoted as Re-MOF-NH 2 (33%), is incorporated. Moreover, its activity is proven to be 3-folds higher than that of Re-MOF with no -NH 2 groups (Fig. 3a). In the absence of CO 2 under a He atmosphere or with no ReTC, no CO generation was observed. The high activity was maintained for at least 6 hours while the molecular catalyst of H 2 ReTC was found to be deactivated within 1 hour 9, 24 , attributed to both ReTC and -NH 2 functional groups covalently bound to the Re-MOF-NH 2 (X%) framework. This signals that it prevents the prevailing deactivation pathway of dimerization commonly observed in photoactive molecular complexes. The stability of the Re-MOF-NH 2 (33%) sample was also tested after the photocatalytic reaction. The PXRD, SEM and TEM analyses (Figs 3c and S9 in the Supplementary Information) show that their crystallinity and morphology are well maintained after the photocatalytic reaction. In addition, the IR spectra of Re-MOF-NH 2 (33%) before and after the reaction (Fig. 3d) demonstrate that conjugation of ReTC and -NH 2 functional groups preserves their chemical configurations after photocatalysis. This is a clear evidence that Re-MOF-NH 2 (33%) enables to give a high conversion rate with the 100% selectivity for conversion of CO 2 into CO under visible light while maintaining its crystal structure, morphology and chemical functionality after the photocatalytic reaction. EXAFS analysis to determine the cooperative mechanism for photocatalytic conversion. The photocatalytic activity in Re-MOF-NH 2 (33%) should be related to the interaction between the ReTC and its neighbouring -NH 2 functional group. The EXAFS analysis, which was performed for Re-MOF, Re-MOF-NH 2 (33%), and Re-MOF-NH 2 (80%) at the Pohang Accelerator Laboratory (PAL, Republic of Korea) on the 7D beamline at 3 GeV energy and the Re L 3 -edge spectra (E 0 = 10207 eV), were also performed in the transmission mode 25,26 to measure the bonding length between the Re metal complex centre and the neighbouring atoms (Fig. 4a). The spectra were obtained through the Athena software based on the IFEFFIT library. The results demonstrate that Re-MOF without any -NH 2 functional group gives two bonding lengths of 1.8 Å and 2.3 Å, corresponding to the bonds for Re-C and Re-Cl in ReTC, respectively. When the -NH 2 functional groups are introduced into Re-MOF-NH 2 (33%) and Re-MOF-NH 2 (80%), we found that the bond corresponding to Re-C was divided into two bonding lengths of 1.4 Å and 2.3 Å. This is because the configuration of carbonyl groups in ReTC becomes asymmetric upon the interaction with the -NH 2 functional groups. Figure 4b gives the clear evidence that the distance between the Re centre and each carbonyl group becomes different. This asymmetric configuration of the carbonyl groups leads to the uneven electron distribution in ReTC for Re-MOF-NH 2 (X%), which would contribute to polarization of ReTC and thus increase the partial electron affinity to make better chance converting CO 2 to CO when carbamate is formed upon interaction with the -NH 2 functional groups.

IR spectroscopy of Re-MOF-NH 2 (X%) in the presence of CO 2 .
The results via the EXAFS analysis could not explain why Re-MOF-NH 2 (33%) gives a higher activity than Re-MOF-NH 2 (80%). In this reason, we have investigated further the interaction between the -NH 2 functional group and CO 2 using the in-situ IR spectroscopy, where each sample of Re-MOF, Re-MOF-NH 2 (33%), and Re-MOF-NH 2 (80%) was kept inside the chamber and the IR spectra were measured with the chamber under vacuum or filled with CO 2 . The IR spectra of Re-MOF-NH 2 (33%) are shown in Fig. 4c and those of Re-MOF and Re-MOF-NH 2 (80%) are shown in Fig. S10a,b in the Supplementary Information. Figure 4d also show the difference spectra of each sample in the presence of CO 2 . The results give that differences in the spectra are evident only in Re-MOF-NH 2 (33%) and that the most pronounced peaks are at 1596 and 1420 cm −1 corresponding to the asymmetric and symmetric COOstretching vibrations, respectively 27 . This shows that amine functional groups helped to result in the intermolecular stabilization of carbamate as shown in Fig. 4e. Theoretical investigation for photocatalytic cooperative conversion mechanism of CO 2 into CO. Figure 5a shows the stable CO 2 -to-CO photochemical reduction pathway determined by using density functional theory (DFT) calculations. The first reaction corresponding to step i occurs via formation of the carbamate complex that is initiated by adsorption of CO 2 into the -NH 2 ligand, where the OH of the carboxylic group forms a week hydrogen bonding interaction with the Cl attached in a neighboring ReTC ligand (Fig. S11a,b).
Steps ii and iii in Fig. 5a give clear evidences that the electron uptake from TEA spontaneously occurs after the photo-excitation process. Due to the additional electron injected in the ReTC, the Re center attacks the electrophilic carbonyl carbon of a nearby -COOH accompanying with the protonation of amine group where the proton is from TEA + . This leads to the migration of the -COOH group to the ReTC ligand along with the regeneration of -NH 2 group (step iv). It is notable that the Re center then forms a 7-coordinate system (Fig. S11c). Subsequently, upon the second photo-excitation process (step v) followed by the second electron and proton uptake from TEA (step vi and vii), the -COOH group is transformed to the final product of CO and the byproduct of H 2 O at the ReTC ligand. Consequently, these support that the formation of the carbamate intermediate in the presence of the -NH 2 ligand is responsible to give efficient production of CO from CO 2 in Re-MOF-NH 2 via the synergistic interaction of two different ReTC and -NH 2 ligands that is not available in Re-MOF having no adjacent amine moiety. Furthermore, the corresponding energy diagram from step i to viii (Fig. 5b) gives more clarifications that the production of CO upon absorption of two photons in visible light is energetically favorable in the presence of the -NH 2 ligand and the ReTC ligand in the neighboring sites. It is also notable that the structural details involving in the photocatalytic reaction have been summarized in Fig. S12. On summary, we designed the chemical environment of a molecular photocatalyst covalently bound to the MOFs through incorporation of the -NH 2 functionalized ligand resulting in a conjugated structure of the neighbouring Re metal complex ligand. The ratio of the -NH 2 functional groups was varied from 0 to 80 mol% in Re-MOF, and the results showed that the photocatalytic CO 2 to CO conversion reached the highest activity when 33 mol% of the -NH 2 functional groups was incorporated. In addition, the EXAFS and IR spectra supported the cooperative reaction mechanism in Re-MOF-NH 2 (33%). These results revealed that the -NH 2 functional groups induced the different bond lengths for Re-CO in ReTC while they help to make the intermolecular stabilization of carbamate with CO 2 , thus increasing its photocatalytic activity.

Methods
Synthesis of Re-MOF-NH 2 (X%). We have prepared the ligand and metal units for the MOF separately.
First, ZrCl 4 (9.3 mg, 0.04 mmol) was dissolved in N,N'-dimethylformamide (DMF, 5 mL) with acetic acid (0.5 mL) in a 20 mL glass vial. The ligand was prepared using a solution of rhenium ligands (4.4 mg, 0.008 mmol), H 2 -BPDC-(NH 2 ) 2 (0, 20, 40, 60, and 80 mol% of H 2 ReTC), biphenyl-4,4′-dicarboxylic acid (adding up to a total amount of 0.040 mmol of organic linker) and DMF (5 mL). Then, the well-dispersed ligand solution was poured into the metal solution, and the sealed vial was placed in an 85 °C oven for 12 hours without light. Then, it was cooled to room temperature and the solid was separated with a centrifuge (8000 rpm for 10 minutes). Next, it was washed with DMF once and methanol three times. After washing, the product was dried in a vacuum oven.
Photocatalytic measurements. The photocatalytic CO 2 conversion experiments were conducted using a stainless-steel reactor. To accomplish this, 5 mg of the sample was placed on a glass holder and the reactor was fastened with screws. To remove solvent and air, the reactor was kept under vacuum overnight. Then, the reactor was purged using CO 2 gas (99.9%) for 30 minutes. Next, 0.1 mL of trimethylamine (TEA) was injected into the chamber, which was kept at 40 °C for approximately 15 minutes to vaporize the TEA. The sample was irradiated through the quartz glass using a 300 W Xenon arc lamp (ORIEL) equipped with an IR water filter and a UV filter to obtain visible light ranging from 400 to 700 nm. The evolved gas was sampled using an airtight sample-lock syringe (Hamilton, 81256) and was injected into a gas chromatography column (Shimadzu, GC-2014A) and passed through a HayeSep Q Porapak Q column with helium carrier gas. The gas was detected using a flame ionization detector (FID) and a methanizer.

EXAFS analysis.
The extended X-ray absorption fine structure (EXAFS) analysis was conducted at the Pohang Accelerator Laboratory (PAL) on the 7D beamline at 3 GeV energy. A Si(111) double crystal monochromator was used to monochromatize the synchrotron radiation and the Re L3-edge spectra (E 0 = 10207 eV) were obtained in transmission mode. To minimize contamination of higher harmonics, the incident beam was detuned by 10%. Also, the intensity of the beam was monitored by using a N 2 -filled IC SPEC ionization chamber. This experiment was conducted with 0.4 eV steps between 10515 and 10585 eV, 30 eV steps between 10585 and 11707 eV, and 50 eV steps between 11707 and 12007 eV with 2, 3, and 3 seconds per the point, respectively. Then, the spectra were obtained through the Athena software based on the IFEFFIT library.
In-situ IR Spectroscopy. The in-situ IR spectroscopy was measured by the FT-IR spectrometer equipped with a gas chamber. The Re-MOF-NH 2 (X%) samples were diluted with the KBr powder and placed in the sample holder. Then, before measuring the IR spectrum in vacuum, the chamber was evacuated for 30 minutes. Next, the chamber was filled with CO 2 gas for 30 minutes to measure the IR signal under CO 2 atmosphere. Each sample was measured with a 4 cm −1 spectral resolution and a 4 mm/cm scanning speed.
Computational Details. DFT calculations were performed using the Jaguar 8.9 software 28 for theoretical reaction energetics. We used the range-separated exchange-correlation functional of CAM-B3LYP 29 the reliable description of band edge positions, and used the LACVP basis set 30 . Also, the ground electronic and geometric structures were fully optimized under constraints with the fixed coordinates of oxygen atoms in terminal carboxylic functional groups from experimentally observed atomic positions. For the steps involving electron/proton transfer, the electron transfer is determined on consideration of coupling with the oxidation of TEA corresponding to the reaction of TEA → TEA + + −e − while the proton is transferred from the alpha carbon of TEA + . The photo-excitation energy is calculated using the HOMO-LUMO gap.